A young startup that emerged from stealth in less than two months is back in the spotlight with a high-profile backer and bigger ambitions to make it big in the world of AI.
Decart is building what CEO and co-founder Dean Leitersdorf describes as a “fully vertically integrated AI research lab” alongside enterprise and consumer products based on the lab's work. I am. The first enterprise product to optimize GPU usage has already generated millions of dollars in revenue. And its first consumer product, a playable “open world” AI model called Oasis, was released when Descart came out of stealth and has already attracted “millions” of players. Masu.
Now, on the back of that strong exit, Descart has raised an additional $32 million led by Benchmark.
The Series A funding comes less than two months after the San Francisco-based company, which has substantial operations in Israel, raised a $21 million seed round from Sequoia and Zeev Ventures. Both companies are participating. In this latest Series A,
And TechCrunch understands from sources that Descart's new post-money valuation is now more than $500 million. (For comparison, Seed valued it at just over $100 million.)
Reitersdorf is a youthful 26-year-old, full of energy and growing quickly. The goal, he said, is not just to compete with companies that are already known as big players in the AI space, such as OpenAI, Anthropic and Mistral. He said he wants to build a “kilocone,” or a $1 trillion company.
“We still have a long way to go and great things to build,” he added. That said, he noted that the company has already been approached as an acquisition target multiple times. And if you just look at the optimization part that Descart has built, there are some interesting (if somewhat modest) comparisons, including Run:ai, which was acquired by Nvidia for $700 million.
Still, Reitersdorf's exuberance comes after a decade of remarkable momentum that led him to his current position.
Reitersdorf was born in Israel and spent his childhood there before following his parents' careers (physicians and researchers) with his family and moving to Switzerland and then Palo Alto.
As a teenager attending Palo Alto High School, he hustled to earn his diploma in just two years, but then made the jump to college, returning to the Technion in Israel to complete his undergraduate, master's, and Ph.D. degrees in computer science. It took only two years to complete the acquisition. Five years, including the period that overlaps with military service.
His co-founder Moshe Shareef (pictured above, left) is impressive in another way. He grew up in a strict Orthodox family and came to the world of computer science while serving his time in the IDF. He proved to be talented and helped establish, build, and execute AI operations for the IDF's 8200 Intelligence Unit, where he remained for nearly 14 years.
There is a third co-founder with an equally impressive resume, but his name has not yet been revealed due to existing commitments.
Decart before the horse
As Leitersdorf explains, Decart, as it currently exists, focuses on three main areas: systems (currently: infrastructure optimization), models (AI algorithms), and data (applications that ingest and return data). I am.
Decart's first product actually launched earlier this year, still in stealth, and is included in the systems camp. This is software that helps optimize how GPU processes operate when training and running inference workloads on AI models.
This software has been found to work very well and is used by many companies that build and run models to reduce some of the extreme operational costs associated with building and using artificial intelligence. . Leitersdorf said the company's software can reduce workloads that typically cost $100/hour to run to just 25 cents/hour.
“That definitely got people’s attention,” he joked. Yes, AI is getting a lot of attention right now, but the companies that are developing technology to improve its capabilities seem to be getting even more attention.
The company hasn't disclosed the names of its customers, but it has already generated millions of dollars in revenue and has enough customers that Descart was profitable when it launched in early November. It is claimed that there is. Reitersdorf said the company is currently on track to remain profitable through the end of the year, adding that interest from the market could be another reason for VC interest.
“Decart’s innovations are making AI generation not only more efficient, but also more accessible to all types of users,” said Victor Lazarte, general partner at Benchmark, who led the deal. stated in a statement. “By removing barriers to entry and significantly reducing costs, we are driving a new wave of creativity and practical application. We hope they will reimagine the potential of AI and its role in our daily lives. I am proud to be a part of this defining journey.”
So far, its optimization product has not been Decart's main focus, although it may be the driving force behind the startup's profits. Reitersdorf said Descartes built it to help finance his business while it was still in stealth mode, based in part on research he did as a student.
Reitersdorf said Descart's second product is in keeping with what the company hopes to do more of in the future.
The Minecraft-like Oasis, which the company launched upon coming out of stealth two months ago, is a “playable” AI that generates responsive AI-based audio and visual interactions in real-time.
The plan, Reitersdorf said, is to launch more experiences along these lines. These include the upgraded Oasis game, as well as other games that utilize generative AI and interactivity. These may include AR or VR experiences that do not require specific hardware to run.
“problem [with VR and AR previously] “We started with hardware rails,” he said. “But hardware is hard to build, and it's hard to get people to adopt new hardware. The good thing about Gen AI is that you can actually: [build AR] In the software part. In fact, it can deliver value even before the hardware is ready. ”
Some would argue that Deccart, ironically, may be putting the cart before the horse when it comes to some of its ambitions. Leitersdorf didn't give me much answers about the company's position on customers who want to use the optimization software to build or run fraudulent models.
Additionally, the company currently has no plans in place to ensure that the applications it develops are not exploited. At this point, he said, those scenarios are not realistic.
More importantly, get more people interested in your work across platforms and turn that activity into revenue.
“The real king is the user,” Reitersdorf said. “They're the only ones that matter.”